AI-powered flash flood alert mini-program launched in SW China province

CHENGDU -- An AI-powered precision early warning system for flash floods and debris flows in the mountainous Sichuan province in southwestern China has been officially launched to the public via a WeChat mini-program, offering real-time risk monitoring and alerts.

Developed by the Institute of Mountain Hazards and Environment (IMHE) under the Chinese Academy of Sciences, based in Chengdu, capital of Sichuan, the system integrates multi-departmental real-time data, including geological surveys and meteorological observations, across all flood-prone ravines in the province, Science and Technology Daily reported on Thursday.

It employs high-precision rainfall forecasting models to provide dynamic short-term predictions covering the past hour and the following two hours, significantly improving forecast accuracy ahead of the flood season from late July to early August, according to the report.

The mini-program generates 24-hour, 12-hour and 2-hour risk maps, delivering warnings via SMS, WeChat and the emergency response systems.

"People can check live monitoring footage, rainfall data, forecasts and hazard warnings to make informed evacuation decisions," said Ouyang Chaojun, an IMHE researcher.

"They can also upload photos of emerging risks to help local governments evaluate the risks and respond," Ouyang added.

Powered by efficient algorithms and the National Supercomputing Center in Chengdu, the system significantly improves the accuracy of rainfall predictions and the efficiency of disaster simulations in mountainous areas.

Since its trial launch in May in the city of Mianyang, where it monitors all 676 ravines, the system has issued 2,034 warnings, successfully forecasting debris flows and flash floods.